GLCM-based Texture Features and Artificial Neural Network (ANN) for Accurate Detection of Melanoma and Basal Cell Carcinoma

Existing studies on image-based detection of melanoma and Basal Cell Carcinoma (BCC) using artificial intelligence (AI) have shown suboptimal detection results due to limitations in feature extraction techniques and classifier algorithms. This research addresses these challenges by developing an Artificial Neural Network (ANN) model that utilizes Gray Level Co-occurrence Matrix (GLCM) features. Dermoscopic images of melanoma, BCC, and normal skin were processed to extract texture features using GLCM. The performance of the proposed model was evaluated using key metrics, including accuracy, sensitivity, specificity, and F1-score. The results demonstrate that the ANN model incorporating GLCM features exhibits strong potential for accurately classifying melanoma, BCC, and normal skin. During the training phase (GLCM with d=2), the model achieved an average performance of 83.33% accuracy, 83.07% sensitivity, 91.54% specificity, and 83.10% F1-score. On the testing dataset (GLCM with d=5), the model achieved 82.22% accuracy, 82.22% sensitivity, 91.11% specificity, and 82.22% F1-score, respectively. These findings highlight the effectiveness of GLCM features in enhancing the ANN model's classification performance and its potential application in clinical settings.